Xo. 641] SHORTEI! AUTK U-.s AM) C O IlUEs P< ,\ DKXCE 565 



gressive variation in F, would not come witliin the scope of 

 blending inheritance as I defined it, when proposing the method, 

 but only eases in which both and are intermediate between 

 the parent races. But Dr. Shull maintains that some dominant 

 factors may be involved even where dominant characters are not 

 in evidence. This may be admitted as a possibility even though 

 we have no evidence for it. What of it 1 Dr. Shull formulates 

 a test case, the strongest one imaginable, in which all (six) 

 factors affecting a size character are "completely dominant" 

 and finds that in this case the method which I suggested would 

 indicate a smaller number of factors than the true one, or four 

 instead of six. If this is the maximum error to be anticipated 

 when all factors are completely dominant, and it is really doubt- 

 ful whether any factors are dominant in the ease of blending 

 characters, the possibility need not give us great concern. 

 Further, if all factors involvi^l in pi-oduciiig a character are 

 ''completely dominant,'' how cini ilic chjii-jicter itself keep from 

 being dominant? And if it is cloniinant. tlie ease will be auto- 

 matically removed from the field of blending inheritance. Shull 

 seeks to avoid the difficulty in his hypothetical case by putting 

 three dominant factors in one luii-ent race and three in the other, 

 but this arrangement, by recoiubiiiat ion of factors, in F., wouM 



The supposed difficully, that some tailors iiivolvnl in the pnt- 

 duction of blending characters may b'' pn>invf m adion while 

 others arc negative, is purely formal. With ihr. r [)Msitive and 

 three negative factors, in his hypothe) iml ra-.'. coines out 



with identically the same distribution m thir'tm, ^i/c rlasses 

 that 1 calculated for the same number of t'ac-lors all lursitiv.', 

 both of us assuming no dominance to (»>M'nr. 



